In my experience, studying finance books is almost a complete waste of time. There are several platforms, such as Interactive Brokers, that provide APIs and market data feeds that you can get started on writing trading programs very quickly. Don't think that you are going to invent profitable trading strategies, at least not for a long time. There are lots of interesting and useful exercises that are lots of fun and can be executed with very little capital, for example, try building an algorithm that executes N shares of stock over the day and executes them at an average price that gets as close as possible to the volume weighted average price over the day. If you can do that, you'll have learned as much as most working algo developers, and if you're trying to break into a job in finance, you'll have a big advantage over most candidates who have dabbled because you will be about the only one who has built something useful rather than focusing on some hokey alpha signal.
If you really want to trade your own money for profit, yes, the space is crowded and HFTs have already found most of the profitable signals that scale, but it is still possible to find profitable signals if you source data that is somewhat new or unique and you look for signals that are capacity constrained and therefore of less interest to HFTs.